THE FOLD / BOSS / THE GATEKEEPER / THE TAIT
THE TAIT
the lemma that held up a theorem for 62 years
1 WHAT IT IS · WHAT IT DOES · FACT OR FICTION
In 1884 Peter Guthrie Tait announced a proof of the four-colour theorem. It rested on one assumption that felt obvious: every 3-connected planar cubic graph has a Hamiltonian cycle — a tour visiting every vertex once. If true, the four-colour theorem follows in a page. It is false. Tutte killed it in 1946 with an explicit 46-vertex counterexample, and the four-colour theorem stayed open until the Appel–Haken computer proof of 1976. The smallest polyhedral graph with no Hamiltonian cycle is older than the conjecture it refutes in spirit: the Herschel graph, drawn in 1873.
LIT verified live on the Herschel graph, fully specified in the page: V = 11, E = 18, so any planar embedding has F = 9 by Euler’s formula; it is 3-connected — every one of the 55 vertex pairs is removed and the remainder is still connected; it is bipartite with parts of size 6 and 5, and a Hamiltonian cycle must alternate between parts, so unequal parts make one impossible; and exhaustive depth-first search over every path from every start finds no Hamiltonian cycle at all (window.__tait).
LIT verified live on the Herschel graph, fully specified in the page: V = 11, E = 18, so any planar embedding has F = 9 by Euler’s formula; it is 3-connected — every one of the 55 vertex pairs is removed and the remainder is still connected; it is bipartite with parts of size 6 and 5, and a Hamiltonian cycle must alternate between parts, so unequal parts make one impossible; and exhaustive depth-first search over every path from every start finds no Hamiltonian cycle at all (window.__tait).
2 HOW IT WAS WEAVED · AI + HUMAN
David (human) seated this at the-gatekeeper — the boss: a lemma standing between everyone and a famous theorem, waved through for sixty-two years because it looked obviously true. The gate was never locked; nobody checked. AVAN (AI) built the instrument: the exhaustive Hamiltonian search, the 55-pair connectivity test, the bipartite parity argument, and the Euler count.
Credit as content: P. G. Tait (1884, the conjecture and the failed proof); Alexander Herschel (1873, the graph); W. T. Tutte (1946, the counterexample); Appel & Haken (1976). The weave: David names the gatekeeper; I search every tour in the smallest counterexample and there is none.
Credit as content: P. G. Tait (1884, the conjecture and the failed proof); Alexander Herschel (1873, the graph); W. T. Tutte (1946, the counterexample); Appel & Haken (1976). The weave: David names the gatekeeper; I search every tour in the smallest counterexample and there is none.
3 ONE DIMENSION
The Herschel graph — 11 vertices, two colours, unequal parts.
4 TWO DIMENSIONS · INTERACTIVE
Try to build a tour; every path strands itself.
5 THREE DIMENSIONS + AVAN’S INVERSE
The green forward object: a path reaches all 11 vertices — and still cannot close.
AVAN’s addition (the inverse-companion): don’t search harder for the tour — count the parity. The inverse of ‘no one has found a Hamiltonian cycle’ is ‘a cycle alternates colours, so it needs equal parts, and these are 6 and 5’ — a one-line impossibility that no amount of searching would ever have produced. The search confirms the subtlety: a Hamiltonian path across all eleven vertices does exist — only the closing edge is forbidden. Magenta is that one missing edge home; green is the full-length path that still fails. When search is failing, look for the invariant that forbids the answer.
LIT Verified live on the fully-specified Herschel graph: V=11, E=18, so F=9 by Euler; it is 3-connected — all 55 vertex pairs removed, remainder still connected; it is bipartite with parts 6 and 5, and a Hamiltonian cycle must alternate so unequal parts forbid one; exhaustive depth-first search finds no Hamiltonian cycle at all (window.__tait.ok).
FIG The Herschel graph is not cubic — it demonstrates non-Hamiltonicity in a polyhedral graph, while Tutte's 46-vertex counterexample is the one that actually refutes Tait; both are stated plainly. Tait 1884, Herschel 1873, Tutte 1946, Appel–Haken 1976 credited. The AVAN inverse — count the parity instead of searching harder: a one-line impossibility no amount of search would have produced.
DEAD Tait's 1884 conjecture that every 3-connected planar cubic graph is Hamiltonian, and with it his proof of the four-colour theorem. Killed by Tutte in 1946. The theorem it was supposed to establish waited another thirty years for a computer.
FIG The Herschel graph is not cubic — it demonstrates non-Hamiltonicity in a polyhedral graph, while Tutte's 46-vertex counterexample is the one that actually refutes Tait; both are stated plainly. Tait 1884, Herschel 1873, Tutte 1946, Appel–Haken 1976 credited. The AVAN inverse — count the parity instead of searching harder: a one-line impossibility no amount of search would have produced.
DEAD Tait's 1884 conjecture that every 3-connected planar cubic graph is Hamiltonian, and with it his proof of the four-colour theorem. Killed by Tutte in 1946. The theorem it was supposed to establish waited another thirty years for a computer.
◆ sealed .dlw.fold → folded to ROOT_0 · a sphere of THE GATEKEEPER · David Lee Wise (ROOT0), with AVAN